Tom Lane escribió:
> Peter Eisentraut <peter_e@gmx.net> writes:
> > On 1/23/14, 4:08 PM, Robert Haas wrote:
> >> Why wouldn't that be necessary with your approach, too? I mean, if
> >> there's a GUC that controls the event source name, then it can be
> >> changed between restarts, regardless of what you call it.
>
> > I don't know if it's practical, but the logical conclusion here would be
> > to use an identifier that you cannot change, such as the system identifier.
>
> That particular ID would be a horrid choice, because we don't try very
> hard to ensure it's unique. In particular, a standby server on the same
> machine as the master (not an uncommon case, at least for testing
> purposes) would be a guaranteed fail with that approach.
>
> I'm still not clear on why we can't just use the port number.
I wonder if it would make sense to generate a unique name at some
initial point in the history of the service (perhaps at initdb time, or
at the first postmaster start) and store this name in a special,
separate file in PGDATA. On subsequent starts we read the name from
there and always use it consistently.
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services